Are SME Loans Less Risky than Regulatory Capital Requirements Suggest?

Transcription

1 Are SME Loans Less Risky than Regulatory Capital Requirements Suggest? Klaus Düllmann Philipp Koziol EBA Research Workshop, London 14 November 2013 Deutsche Bundesbank This paper represents the authors personal opinions and does not necessarily reflect the views of the Deutsche Bundesbank or its staff.

2 Motivation and Contribution Higher minimum capital ratios and a tighter capital definition in Basel III have indirectly also affected capital requirements for credit exposures to SMEs Do these regulatory adjustments treat SME unfairly given they didn t cause the recent financial crises? Empirical literature is inconclusive but tendency towards lower asset correlation estimates than those in the corporate risk weight functions of Basel II Contribution: 1. Assess the systematic risk of German SME loans measured by the asset correlation in a common asset value credit risk model 2. Compare estimation results with capital requirements for SME lending under the CRR / CRD 4 framework 3. Unique data sample of SME lending by over 400 small and large German banks Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 1

3 Contribution and Overview Step 1: Estimate asset correlations (AC) from historical default rates of selected size and rating buckets Step 2: Compare the size-dependence of IRB risk-weights with the sizedependence of empirical risk-weights (i.e. risk weights based on estimates of AC and PD) Focus on relative calibration : Does the regulatory capital for SMEs appropriately reflect the systematic risk relative to other asset classes? Use IRB capital requirements (based on the asymptotic single risk factor model) and not asset correlation estimates directly for a comparison because they are the economically relevant measure Large corporates serve as benchmark, i.e. we assume that their IRB risk weights are correctly calibrated Carry out various robustness checks for estimation results Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 2

4 Data Data on more than 400 German banks (both small and large banks) Default rates in the credit portfolio Borrowers: domestic firms except for credit institutions with available IRB PDs (no retail and specialized lending) Number of borrowers as of the beginning of each period Number of defaults occurring during the period under consideration Data clustering of default rates along three dimensions: 1. Time period: 14 semi-annual periods, 1 June 2005 to 31 December 2011 (7 years), seasonally adjusted 2. Rating category: Six rating classes based on IFD master scale aggregated: I III, IV, V VI 3. Size: Measured by yearly turnover (in m e ): [0, 0.3], (0.3, 1], (1, 2.5], (2.5, 5], (5, 50], > 50 Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 3

5 Data number of ratings and defaults (a) # Ratings by rating category (b) # Defaults by rating category Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 4

6 Data default rates Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 5

7 Data default rates Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 6

8 Model and estimation methodology Ability-to-pay process of firm i: Y i = ρx + 1 ρε i systematic risk-factor X N(0, 1), idiosyncratic risk-factor ε i N(0, 1), asset correlation ρ Conditional default probability: P (L = 1 X = x) = Φ Estimation technique: ( ) γ ρ x 1 ρ Maximum-Likelihood (ML) estimator by Gordy & Heitfield (2002), used for Basel II calibration, downward bias for small samples Robustness checks through Method-of-Moments (MM) and Asymptotic Maximum Likelihood (AML) without bias correction, yearly estimations... Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 7

9 Model and estimation methodology Empirical risk-weight formula: RW (LGD,PD,M, ρ) = LGD [ Φ ( Φ 1 (PD) + ) ] ρφ 1 (0.999) PD f(pd,m) 1 ρ Basel II risk-weight formula: RW (LGD,PD,M) = LGD [ Φ ( Φ 1 (PD) + ) ] ρ(pd,s)φ 1 (0.999) PD f(pd,m) 1 ρ(pd,s) where ρ(pd,s) = 1 e 50PD e 50 ( 1 1 e 50PD 1 e 50 ) ( 1 Other retail: turnover < 2.5 m e ; S:= turnover; M:= maturity ) min{50, max{s, 5}} 5 45 Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 8

10 Results Risk weights per rating and size class Other Retail Corporate Estimates Turnover [0, 0.3] (0.3, 1] (1, 2.5] (2.5, 5] (5, 50] > 50 Rating I-III IV V-VI Other Retail Corporate Basel II Turnover [0, 0.3] (0.3, 1] (1, 2.5] (2.5, 5] (5, 50] > 50 Rating I-III IV V-VI Est,V V I Relative difference for estimated RW: 5 50 BII,V V I Relative difference for Basel II RW: 5 50 = = 28.5 = = 15.0 Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 9

11 Results Relative differences by rating and turnover class Other Retail Corporate Estimates Turnover [0, 0.3] (0.3, 1] (1, 2.5] (2.5, 5] (5, 50] > 50 Rating I-III IV V-VI Other Retail Corporate Basel II Turnover [0, 0.3] (0.3, 1] (1, 2.5] (2.5, 5] (5, 50] > 50 Rating I-III IV V-VI Reductions are calculated as a weighted average with respect to the number of loans per rating class Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 10

12 Results analysis of IRB and RSA risk weights IRBA Other Retail Corporate Turnover [0, 0.3] (0.3, 1] (1, 2.5] (2.5, 5] (5, 50] > 50 Differences Basel II IRBA -49.3% -50.2% -48.9% -13.3% -10.3% 0.0% Estimated -42.7% -47.4% % -35.1% -33.9% 0.0% Total Difference 6.6% 2.8% 9.2% -21.8% -23.6% 0.0% RSA Other Retail Corporate Turnover [0, 0.3] (0.3, 1] (1, 2.5] (2.5, 5] (5, 50] > 50 Differences Basel II RSA -25.0% -25.0% -25.0% 0.0% 0.0% 0.0% Estimated -42.7% -47.4% -39.7% -35.1% -33.9% 0.0% Total Difference -17.7% -22.4% -14.7% -35.1% -33.9% 0.0% Total differences are averages over rating categories. Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 11

13 Summary Consider total differences > 10% between Basel II and estimated risk weights as economically significant Then total differences are significant for SMEs in the IRB corporate portfolio (annual turnover between 5 and 40 mln EUR) generally under RSA Before drawing policy conclusions the following caveats should be considered Basel is an international framework; results for other countries necessary before risk weights functions should be revisited (work in progress). RSA was calibrated more conservatively than the IRBA since it is much less risk sensitive. This can at least partly explain significant total differences. Time series of default rates is till relatively short and may not cover a representative economic cycle. Klaus Duellmann and Philipp Koziol, Deutsche Bundesbank 12